Three hundred million dollars to put a world model in front of Amazon's chips — with Nvidia on the cap table. That can be visionary, or it can be a conflict of interest in a nice suit. This is Startup Fundraising. Today we've got Radical Ventures leading the week's biggest round, Sequoia riding along — and the real question is who actually gets to set the price anymore. Also on the board: a $400M fintech fund where Santander is the only LP, and a $25M defence round that is, frankly, the easiest use-of-funds to understand today. One of those I can actually audit. Decart first — because when a company raises four hundred and fifty million dollars in two years, I want the cap table before I want the press release. The Next Web, with Ana-Maria Stanciuc:
Decart, the AI research lab building real-time video and world models, announced on Monday it had raised $300m in new funding led by Radical Ventures. The round takes the two-year-old company’s total raised past $450m, with Nvidia, Atreides Management, Valor Equity Partners, Adobe Ventures, Toyota Ventures, and eBay Ventures joining as new investors alongside returning backers Sequoia Capital, Benchmark, and Zeev Ventures.
Radical Ventures led this one, not Sequoia, not Benchmark, even though both of them are in there as returning backers. Yesterday Sequoia co-led Dust's Series B; today they're in the passenger seat on a $300M round set by a firm most people outside Toronto would have to point to on a map. Nvidia is a new investor in a company whose headline product runs in front of Amazon's chips. That doesn't read like a financial bet to me — it reads like Nvidia grabbing a window seat to see what the competition is doing at inference time. And they get board information rights on top of that. The angel list is doing real work on the story here — Karpathy, Michael Eisner, the Nintendo family, a gaming investor. That's not random. It tees this up as media, gaming, and infrastructure all at once, which is exactly the kind of framing a pitch deck loves. $450M raised, two years old, and the revenue line in this announcement is about as visible as it was in Wispr's round last week — meaning not visible at all. Karpathy's name is doing a lot of the selling where a deployment number should be. Here's SiliconANGLE:
Dust, an agentic artificial intelligence startup that’s trying to push enterprise workers away from isolated chatbots into a more collaborative, multiplayer ecosystem, said today it has raised $40 million in a Series B round of funding.
Dust's $40M Series B is the follow-on to yesterday's thread — Abstract and Sequoia co-leading, with Snowflake and Datadog writing checks alongside. That says institutional conviction, sure, but the co-lead structure is the part to watch: Abstract and Sequoia are sharing the top of the cap table, so neither one is clearly driving. Snowflake and Datadog showing up as participants is the tell. Those aren't just financial bets, they're distribution options. Both companies want an agentic layer on top of their data stack, and sixty million total raised is a pretty cheap way to buy a seat. Whether Dust has real enterprise contracts or just enthusiastic pilots is still the unanswered part. And Sequoia turns up here as co-lead, then tomorrow as a follower in Decart's three-hundred-million round — that's the lead-versus-follow question in back-to-back days, and I don't think that's an accident. This one's from SmartCompany:
Melbourne defence tech startup Arkeus has raised $25 million at a $100 million valuation in a Series A round led by QIC Ventures. The round drew new investors R+VC, Folklore Ventures, and Dyne Ventures alongside existing backers Main Sequence Ventures, Salus Ventures and Steve Baxter’s Beaten Zone Venture Partners.
SmartCompany had the Arkeus story this morning — $25M Series A, QIC Ventures leading, $100M post-money valuation. Dealroom is calling that a 7x step-up from seed two years ago, which is the number to keep in mind. And the use of funds is actually readable: a Queensland manufacturing facility, a local team, Australian Defence customers. That's a cap table I can work with. Compare that with Decart's $300M and tell me which story is cleaner. The customer concentration question still matters, though. If Australian Defence is basically the whole revenue base, then a 4x implied multiple on a $25M round only makes sense if those contracts are already signed, not just sitting in procurement. Hyperspectral imaging plus onboard AI for drone false-positive reduction — that's a very specific problem they saw fail in the field in Colombia. I'll take a garage-born defence hardware company with a named enemy over a narrative lab any day. Here's Ivan Mehta at TechCrunch:
But those apps don’t do much for patients, which is why Kin Health is building a notetaker that can transcribe your visits to doctors, parse medical advice, and surface next steps when required. To that end, the startup has raised $9 million in a seed funding round led by Maveron.
Kin Health, nine million seed led by Maveron — patient-side AI notetaker, not clinician-side. TechCrunch had it yesterday. The Menlo Ventures number buried in the story is the one I care about: six hundred million in AI notetaking revenue last year across the category. I flagged that Menlo number because six hundred million is real market size, not a vibe. But Kin is going after the patient, not the doctor, which means no reimbursement path, no enterprise contract, and the app is free. So what exactly does nine million buy them before they have to charge somebody? Also worth flagging: the app is explicitly not HIPAA-certified. They say they adhere to the same standards, but that's a meaningful distinction when you're recording people's oncology appointments. Heidi and Freed sell to clinics, and clinics have budget plus a compliance department. Kin is selling to patients, and patients have neither. Nine million against a six-hundred-million category sounds disciplined until you realize the wedge doesn't touch the side that pays. Europesays writes:
Mouro Capital has closed $400 million for its third fund, with full support from its long-term partner Banco Santander. This brings the fintech-focused venture firm’s total investment commitments to over $1 billion for the first time.
Mouro Capital closes its third fund at $400M, with Banco Santander as the sole LP. So much for the independent-VC framing. When one bank writes every check, what you've got is a captive vehicle with a VC label on the door. Seven investments are already out the door from this fund — including ElevenLabs and Sakana AI — which tells me the money was moving before the press release ever landed. That's either real conviction or Santander buying strategic optionality in AI and calling it venture returns. And that's the same thread as Snowflake and Datadog taking cap table stakes as distribution bets — Santander is doing it from the LP seat. It isn't a straight financial return play. It's a bank buying a window into fintech before those startups become competitors. Thirty to thirty-five companies, with initial checks around eight million — that sounds disciplined on paper. But Sakana AI isn't a fintech company. The minute you're backing AI labs through a fintech-labeled fund, the thesis has already wandered, and Santander either knows it or doesn't care. Got a fundraising question, a story idea, or a correction for us? Send it to startupfundraising at lantern podcasts dot com. We read the inbox, and your notes help shape future briefings.
You'll find links to every story we covered today in the show notes, so if one sparked an idea, it's easy to dig in a little deeper. That's Startup Fundraising for Tuesday, May 19. This is a Lantern Podcast.